Essays on housing and pensions a european perspective

Based on Kemeny’s assumption that housing and pension wealth are substitutes andgenerate a trade-off in the asset accumulation phase of a household until retirement,i.e., housing is overr

Essays on Housing and Pensions

Thomas Müller

A European Perspective

Nico B Rottke · Jan Mutl Hrsg.

Series editors

N B Rottke, Frankfurt a M., Germany

J Mutl, Wiesbaden, Germany

Essays in Real Estate Research

Band 15

elle Forschungsarbeiten der Promovenden der Lehrstühle und Professuren des Real Estate Management Institutes der EBS Business School Forschungs- und Lehrschwerpunkte des Institutes bilden die interdisziplinären Aspekte der Immobilientransaktion sowie die nachhaltige Wertschöpfungskette im Immo-bilienlebenszyklus Die Kapitalmärkte werden als essenzieller Bestandteil der Entwicklung der Immobilienmärkte aufgefasst Die in der Regel empirischen Studien betrachten transaktions- und kapitalmarktnahe Themenbereiche aus dem Blickwinkel der institutionellen Immobiliengewerbe- und -wohnungswirtschaft, wie bspw Finanzierung, Kapitalmarktstruktur, Investition, Risikomanagement, Bewertung, Ökonomie oder Portfoliomanagement, aber auch angewandte The-men wie Corporate Real Estate Management, Projektentwicklung oder Unterne-hmensführung Die ersten 11 Bände der Reihe erschienen bis 2014 auch im Immobilien Manager Verlag, Köln.

The series “Essays in Real Estate Research”, published by Professor Dr Nico B Rottke FRICS and Professor Jan Mutl, Ph.D., includes current research work of doctoral students at the chairs and professorships of the Real Estate Management Institute of EBS Business School The research and teaching focus of the Institute constitutes the interdisciplinary aspects of real estate transactions as well as the sustainable value creation chain within the real estate life cycle The capital mar-kets are regarded as essential components of the development of the real estate markets The mostly empirical studies consider transactional as well as capital market topicsfrom the point of view of the institutional commercial and residen-tial real estate industry, such as finance, capital market structure, investment, risk management, valuation, economics or portfolio management, but also applied topics such as corporate real estate management, real estate development, or lead-ership issues in the property industry The first 11 volumes of the series appeared

up until 2014 in Immobilien Manager Publishing, Cologne, as well

More information about this series at http://www.springer.com/series/13911

EBS Universität für Wirtschaft und Recht

Wiesbaden, Germany

Essays in Real Estate Research

Library of Congress Control Number: 2018965574

Doctoral Thesis, EBS Business School, EBS Universität für Wirtschaft und Recht, Wiesbaden, Germany, 2018

Springer Gabler

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2019

This work is subject to copyright All rights are reserved by the Publisher, whether the whole

or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software,

or by similar or dissimilar methodology now known or hereafter developed

The use of general descriptive names, registered names, trademarks, service marks, etc in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.This Springer Gabler imprint is published by the registered company Springer Fachmedien Wiesbaden GmbH part of Springer Nature

The registered company address is: Abraham-Lincoln-Str 46, 65189 Wiesbaden, Germany

https://doi.org/10.1007/978-3-658-24955-7

To my daughter Lilly, who will inherit our present decisions

Covering topics of relevance and proposing solutions are the key drivers of successfulscientific works in social sciences Our publication series “Essays in Real EstateResearch” welcomes relevant topics that are able to create an impact The presentpublication is highly relevant as it touches upon a subject which concerns us all:How to secure adequate pensions in an aging society.

Since the shift in age structure as the main driver for increasing pension expenditures

is not easily reversible, we will face an enormous challenge when trying to ensure equate pensions for current and future pensioners As most of the European pensionsystems are established as so-called PAYG systems, current workers finance currentpensioners An increasing share of pensioners puts inevitably a greater financialburden on current workers Without further compensation, an aging society will soonsuffer from income losses This effect of the demographic change is well known and itsconsequences long debated, but a satisfying solution is not yet implemented in futurepension policies The loss of adequate pensions for most of the European elderly

ad-is still imminent European governments are struggling with serious solutions astheir largest voter potential consists of pensioners or workers just before retirement,e.g., the large cohort of the baby boomer which will retire in the near future But

if we are not honest with ourselves, accept these future challenges, and implementsound solutions, taken actions can only treat the symptoms instead of contributing

to fundamental changes

Looking at the asset distribution within a society rather than at the income tion when assessing pension adequacy, as this work proposes, is an innovative solution.Assets can be income-producing as well, whether in kind or in cash Focusing on hous-ing as the most valuable asset in a European household portfolio is straightforward.The author shows that income derived from housing can help to alleviate financialdistress in old age Providing access to housing and thus enabling households toaccumulate housing wealth must be a crucial policy goal for all European memberstates It cannot be taken for granted, as inter- and intra-generational differences inhousing wealth show already existing inequalities These result from a progressingsegmentation of the housing market and individual income and borrowing constraints,reasons that can be addressed in policy measures In that, a well-directed housingpolicy is essential for a successful welfare policy

distribu-In his empirical work, the author clearly demonstrates that homeownership is alreadypart of a household’s welfare planning Income from housing is an accepted andapplied supplement for public pension income With more and more pressurizedpension systems ahead, one conclusion is obvious: Future pensioners will have to rely

The present work is able to create an impact by providing concrete policy suggestions

as a result of a sound analysis regarding the importance of housing wealth in theEuropean household portfolio It can give European policy makers food for thoughtand inspire their work It is in their hands to create more sustainable and adequatepension systems

Professor Dr Nico B RottkeProfessor Jan Mutl, PhD

The demographic change in Europe is a mega trend that is formative The longer itsdrivers prevail, the stronger will be its effect on our political, social, and economicallife The resulting change in the age structure of the European population calls forconsequences in many fields Therefore, research on the effects of this mega trend

is especially valuable to our society In combination with proposed solutions it iseven more valuable Adaptations to a changing environment are the task of currentgenerations to secure living conditions of future generations

The aim of this thesis is to contribute to intensified discussions on securing adequatepensions in an asset-based welfare system It puts housing as a central pillar There-fore, it gives an answer to one of the most pressing questions that current generationshave to answer: How to provide adequate retirement income for the elderly?This thesis was only possible with the kind support and help of many individuals Iwould like to extend my sincere thanks to all of them

First, I would like to thank my first supervisor, Prof Dr Nico B Rottke for providing

me with the possibility to write my thesis at the Real Estate Management Institute

of EBS University Without his trust in my capabilities, his academic support and hishelpful advice throughout the process of writing and researching, this thesis wouldnot have been possible I would also like to gratefully thank my second supervisor,Prof Jan Mutl, PhD, for being always approachable for methodological as well ascontent issues during this period Furthermore, I am also indebted to acknowledgethe very valuable comments and suggestions for improvement from Prof Dr JoachimZietz His revisions advanced this thesis a lot

Second, I would like to express my gratitude to EBS University, especially the RealEstate Management Institute (REMI), and all research colleagues for their supportand fruitful discussions in our doctoral seminars I’m also very grateful for thevaluable comments I received during my presentation at the 22nd European RealEstate Society Conference in Istanbul Overall, the lectures and discussions which Iattended and the people I met were a real enrichment for my academic life.Finally, I would like to sincerely thank my parents and my wife for their unfailingsupport and continuous encouragement This accomplishment would not have beenpossible without them

Dr Thomas Müller

1.1 Motivation and Problem Definition 1

1.2 Thesis Structure 3

2 The Trade-off between Housing and Pensions 5 2.1 Introduction 5

2.2 The Trade-off Theory 6

2.3 Analytical Framework 8

2.3.1 Theory 9

2.3.2 Econometric Specification 10

2.4 Data 11

2.5 Regression Results 14

2.6 Conclusion 19

3 Do Europeans Release Their Housing Wealth in Old Age? 21 3.1 Introduction 21

3.2 Life-Cycle Theory and Housing Wealth Decumulation 22

3.3 Empirical Model 24

3.4 Data 25

3.5 Regression Results 28

3.6 International Differences 31

3.7 Conclusion 33

4 Housing as a Pension: Policy Implications for Europe 37 4.1 Introduction 37

4.2 Key Challenges for the European Public Pension Systems 38

4.2.1 Demographic Aging 39

4.2.2 Financial and Economic Crisis 41

4.2.3 Effects on Pension Systems 41

4.3 The Role of Housing for Adequate Income in Old Age 45

4.3.1 Housing as an Income Source in Old Age 45

4.3.2 Perception of Housing as a Pension 48

4.4 Economic Theory and Empirical Evidence 50

List of Figures

2.1 Asset allocation of European households 15

3.1 Plot of marginal effects by subgroups ofP 30

3.2 Plot of marginal effects by country 31

4.1 Projected age structure in the EU 2013-2060 40

4.2 Change in public pension expenditure per GDP and influencing effects in the EU 2013-2060 43

4.3 Change in benefit ratio and gross replacement rate in the EU 2013-2060 44 4.4 Housing pension multiples in Europe 46

B.1 Change in old age dependency ratio in the EU 2013-2060 72

B.2 Home ownership rates by country 72

B.3 Poverty rates among the over-65s before and after the inclusion of imputed rents in household income 73

B.4 Average value of main residence by country 73

2.1 Regression coefficients: European sample 16

2.2 Regression coefficients: single country 18

3.1 Summary statistics 26

3.2 Regression coefficients 28

3.3 Robustness: year groups forP 29

3.4 Robustness: margins ofP for year groups 30

A.1 Summary statistics of non-wealth parameters 66

A.2 Regression coefficients: European sample (full estimation output) 67

A.3 Regression coefficients: single country (full estimation output) 68

A.4 Variable table 69

A.5 Regression coefficients (full estimation output) 70

List of Abbreviations

1.1 Motivation and Problem Definition

Mainstream welfare research has rarely discussed the importance of housing in awelfare state (Delfani, De Deken & Dewilde, 2014) On the other hand, housingresearch is considerably influenced by welfare research (Hoekstra, 2013) The role ofhousing in a welfare state, however, has been debated controversially among housingresearchers This thesis contributes to its clarification

Housing is recognized as a pillar of the welfare state beside pensions, healthcare, andeducation Depending on the approach, it is either called a “wobbly pillar” (Torgersen,1987) or “an essential part of welfare provision” (Delfani et al., 2014) This may seemcontradictory but is, in fact, as Malpass (2008) puts it, “looking at different things

in different ways” Torgersen sees housing as a welfare service with a focus on thepublic or social rented sector characterized by a high state influence and marketde-commodification But the housing sector shows different characteristics compared

to other welfare services Beside its public sector, it has a dominant private sectorinvolvement The private housing market and how it is related to the non-housingservices of the welfare state such as pensions forms the perspective of Delfani et

al Following the second approach, looking at housing as a household asset and itsrelation to its welfare provision, this thesis demonstrates that the private housingasset can be as well a cornerstone of the welfare state, especially by supplementingpension provision

Kemeny (1980) developed a well-known approach to study the relation between thehousing market and the welfare state, focusing on public pensions He assumed avery strong inverse relationship, suggesting that both forms of wealth are substitutes.Thereby, he argued on a country level basis, comparing different institutional settings

In a society with a large and generous welfare state which shares risks collectively,homeownership rates are rather low as pensions are maintaining living standards.The housing sector is dominated by large state interventions and a strong rentalmarket On the other side, in a homeownership prevailing society with a large scale

of commodification of the housing market, welfare provision is minimized Castles(1998) named this relation in a response, the “trade-off between homeownershipand pensions” However, as this position puts housing as a causal variable for thearrangement of a welfare state, it has not been without criticism Malpass (2008)found it “hard to accept that housing has such a profound influence on the formand extent of welfare states” Furthermore, Kemeny (1980) had never tested histheory statistically This gap was later filled by Castles and Ferrera (1996) who foundempirical evidence for Kemeny’s thesis but a weakening relation over time Doling

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2019

T Müller, Essays on Housing and Pensions, Essays in Real Estate Research 15,

https://doi.org/10.1007/978-3-658-24955-7_1

2 1 Introduction

and Horsewood (2011) addressed the causality of the relationship, suggesting that

an increase in housing wealth is followed by reductions in welfare state spending.However, Pierson (1996) argued that “the welfare state has proved far more resilient and far more durable than existing theories of the welfare state would lead one toexpect” Therefore, with still a lot of unanswered questions, the relation betweenhousing and the welfare state constitutes an exciting research field

Based on Kemeny’s assumption that housing and pension wealth are substitutes andgenerate a trade-off in the asset accumulation phase of a household until retirement,i.e., housing is overrepresented in household portfolios with low pension wealth andvice versa, a further question arises: Can housing act as a pension after retirement?Housing wealth is income-generating in two forms: It provides income in kind whenconsuming housing services as shelter or family space and income in cash when sellingthe asset There it acts as a store of wealth which can be released when needed Thetheory of a possible wealth decumulation after retirement is based on the life-cycle hy-pothesis by Modigliani (1986) The hypothesis predicts a hump-shaped asset-wealthcurve with its peak at retirement It anticipates decreasing rates thereafter, assuming

an asset decumulation process Empirical evidence for housing wealth decumulation

in Europe is limited Chiuri and Jappelli (2008) found declining homeownership ratesafter the age of 70, analyzing repeated cross-section data out of 60 microeconomicsurveys in 15 OECD countries Furthermore, differences in market regulations causedcountry-specific homeownership patterns Angelini and Laferrère (2011) argued thathousing equity release depends to a great degree on mortgage market access andfinancial regulations Additionally, the provision of public housing and long term careaccommodations affects release pattern Hence, both conclude that the institutionalsetting is a main influence for housing equity release when trading down

Particularly today, pension supplements as income from housing come to fore asEuropean countries face key challenges regarding the funding of their pension systems

At the same time, the European Union (EU) sets “adequate and sustainable ment incomes for EU citizens now and in the future” as a main policy goal (EuropeanCommission, 2010) But European welfare states are struggling to provide adequatepensions in the future as cut backs are necessary to sustain their affordability TheEuropean Commission and Economic Policy Committee (Ageing Working Group)(2015) project decreasing benefit levels and replacement rates which lead to a decline

retire-in pension retire-incomes for the elderly Accordretire-ing to the trade-off hypothesis, housretire-ingcan be one form of alleviating financial distress for future retirees There is a casefor looking into this relationship from the current perspective of pressurized welfaresystems

including pension wealth into a household’s asset allocation, we are able to calculatethe displacement effect between both forms of wealth.

Chapter 3 focuses on income in cash when analyzing if housing wealth is releasedafter retirement Our empirical analysis on housing wealth release pattern is based

on four connected waves of the SHARE dataset We are able to construct a paneldataset with repeated measures of individual characteristics Previous investigationswere built on cross sections The panel structure allows us to control for unobservedheterogeneity Furthermore, we calculate marginal effects on a country level to discussinternational differences

The insights gained in Chapter 2 and in Chapter 3 are beneficial for policy legislators

as European countries encounter challenges in providing adequate pensions Facinglower benefit levels and decreasing replacement rates, households can supplementtheir public pension with income derived from housing A requirement is, however, anasset-based pension legislation which encourages the accumulation and decumulation

of housing wealth Chapter 4 puts theory into current political and economic contextand derives policy suggestions to encourage housing as a pension

Chapter 5 captures our major research findings It emphasizes their contribution tothe current literature and gives suggestions for further research

cross-national panel database of micro data on health, socio-economic status, and social and family networks of more than 120,000 individuals aged 50+ This thesis uses data from SHARE Waves

1, 2, 4 and 5 (DOIs: 10.6103/SHARE.w1.600, 10.6103/SHARE.w2.600, 10.6103/SHARE.w4.600, 10.6103/SHARE.w5.600), see Börsch-Supan et al (2013) for methodological details The SHARE data collection has been primarily funded by the European Commission through FP5 (QLK6- CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, SHARELIFE: CIT4-CT-2006-028812) and FP7 (SHARE-PREP: N°211909, SHARE-LEAP: N°227822, SHARE M4: N°261982) Additional funding from the German Ministry of Education and Research, the Max Planck Society for the Advancement of Science, the U.S National Institute on Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, HHSN271201300071C) and

2 The Trade-off between Housing and Pensions

The portfolio choice behavior of households may provide some salient answers to thisquestion The household asset portfolio is partly driven by its income expectations.Households form expectations not only about their future labor income, but alsoabout their pension income In states with low public pensions, households mustmake provisions for their old age to supplement any claims on public pensions This

is part of a private redistribution process Individuals devote a part of their laborincome in their working years to private savings so they can secure a certain incomelevel in their old age This is mostly done by buying into private pension schemes,savings accounts, stock market shares, or bonds But the owner-occupied house,which is often the most valuable asset in the household’s portfolio, can also serve as

an insurance policy against old age poverty

Housing acts in a twofold manner: As a consumption and an investment good.Therefore, it offers two forms of income: Income in cash and income in kind (Doling

& Ronald, 2010) Income in cash reflects the equity value of housing as a tradeablegood which can be sold It is a store of wealth, although its value is driven by marketforces and may increase or decrease over time Wealth can be released by sellingthe home for cash, or it can be converted into a continuous cash income throughequity release schemes (ERS) such as reverse mortgages Income in kind representsthe consumption of services delivered by the housing asset Housing services arederived from size, location and the ability of the house to serve as a shelter, familyspace, etc Income in kind is measured as rental value minus any housing costs Forfirst-time buyers, housing costs comprise not only maintenance and service charges.Mortgage payments are a significant part of these costs, which can represent up

to a third of household income (Kemeny, 1980) To the extent that house prices

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2019

T Müller, Essays on Housing and Pensions, Essays in Real Estate Research 15,

https://doi.org/10.1007/978-3-658-24955-7_2

increase rather than fall over time, owners experience a capital gain, which increasestheir net worth Even for constant house prices the net worth rises over time, asthe repayment of mortgage debt advances and owners accumulate equity in theirhomes In their retirement years, homeowners can add income in kind or income incash derived from housing to any public pension income they may receive Hence,income from public pensions and income from housing can be considered altern-ative ways to maintain adequate living standards in old age (Castles & Ferrera, 1996).The ability to substitute one form of income for another creates a trade-off betweenhomeownership and pensions The trade-off theory was introduced by Kemeny (1980)and empirically evaluated by Castles and Ferrera (1996) and Doling and Horsewood(2011) amongst others However, to the best of our knowledge, the theory has neverbeen tested at the household level with comprehensive survey data This analysisadds to the current literature by evaluating the trade-off between pension wealth andhousing wealth using survey data of 11 European countries drawn from SHARE Webase our empirical work on an explicit life-cycle model of the type used in welfaretheory Our empirical results favor the trade-off theory In particular, we estimatethe substitution effect of both wealth forms to be between 16% and 22%.

This chapter is structured as follows First, we outline the trade-off theory and thequalitative and quantitative evidence that has been collected to date Second, wepresent the analytical framework, which covers the underlying life-cycle model andthe econometric specification Third, we detail the data Next, the estimation resultsare presented and discussed We conclude the chapter by summarizing our resultsand suggesting further research

2.2 The Trade-off Theory

The theory of substituting income from public pension and housing in old age, atrade-off between both forms of wealth, was first formulated by Kemeny (1980) Therationale behind this theory is straightforward Countries with a generous welfaresystem and high public pensions force the redistribution of lifetime income to old agethrough, e.g., tax payments In countries with low public pensions, households aredemanded to make their own provisions When households pay off their mortgages,little excess income is available for social transfers funded by taxation Kemeny(1980) reasoned that the relationship between homeownership and welfare must benegative But he did not conduct any statistical tests

Subsequent work by Castles and Ferrera (1996) examined the relationship betweenhomeownership rates and welfare indices in 20 OECD member states The empiricalresults support the hypothesis of an inverse relationship However, they also found

a decline in the relationship over time Furthermore, they questioned the direction

of causality Doling and Horsewood (2011) addressed Castles’ question of causality.Using Granger causality tests, they found statistical support for the idea that anincrease in housing wealth is followed by a decrease in public spending on older

2.2 The Trade-off Theory 7

people More housing equity reduces the need for high public pensions Dewildeand Raeymaeckers (2008) analyzed the issue of old age poverty in a policy context

by contrasting social transfers and housing policies Although they found empiricalsupport for the trade-off hypothesis, they pointed out that not all retirees benefitfrom high homeownership rates In a system with low public pensions and highownership rates, non-owners face a double disadvantage

Delfani et al (2014) suggested that the institutional setting has to be considered whenpromoting a trade-off They stated that if market allocation determines pensions andhousing wealth, then a trade-off is more probable to occur If there is a high degree

of state intervention, both forms of wealth lack a substitution effect Interventions inhousing markets are, e.g., regulations and subsidies for tenants and homeowners Ahigh share of publicly provided pensions and underdeveloped private pension marketsconstitute state interventions in pension provision The trade-off is supposed toprevail only in liberal markets where housing and pensions are exposed to marketrisks and their provision is voluntary If pensions are decommodified, housing is notseen as an investment good that offers income in old age (income in cash) but as

a consumption good that provides housing services (income in kind) Institutionalsettings differ between countries and define the risks to which households are exposed

In their work, Delfani et al (2014) classified countries into four groups of institutionalsettings The first group of Western European countries, which include Germany,France, and Austria, possesses highly regulated pension and housing markets with ahigh degree of rent control In these countries, homeownership is less important as

a pension supplement, as public pensions are adequate for most of the households

A second group, which includes Denmark, the Netherlands, and Sweden, is terized by less housing regulation, but a high portion of mandatory pensions Inthis second group, no asset-based welfare regime is developed, as pensions are notexposed to market forces Nevertheless, countries in this group encourage privatepensions that are funded by assets But as these are often designed as definedbenefit schemes, there is no risk for income shortfalls in old age Countries of thethird group, which include the UK and Ireland, are the only European countrieswith a market-based approach for both housing and pension markets They arecharacterized by a low level of rent control and tenant protection At the same time,public and mandatory private pensions are not sufficient to maintain the level ofincome after retirement Households have to rely on market forces to maintain theirstandard of living and cannot rely on intergenerational solidarity They are forced

charac-to accumulate assets, for example via investments in housing, from which they candraw in old age These countries often have well-developed ERS markets that help torelease equity when necessary Belgium, Portugal, Spain, Italy, and Greece form thelast group They are characterized by underdeveloped rental and mortgage markets,where households are forced into owner-occupation financed by inheritance or familytransfers They face limited coverage by public pensions, which results in a high share

of old age poverty Households in these countries are often characterized as asset richbut income poor Therefore, effects of country-specific interventions in both mar-kets have to be considered when analyzing the trade-off between housing and pensions.

A more qualitative approach on the perception of housing as a pension was ized within the European research project “Demographic Change and HousingWealth” (DEMHOW) Several interviews were conducted by various researchersanalyzing the perception and behavior of households in European countries regarding

coun-tries, Elsinga and Mandi (2010) observed that current retirees see home equity as afinancial reserve and tend to hold on to their home, whereas younger generations donot see the need for a reserve yet However, they concluded that from a qualitativeperspective a trade-off between housing and pension exists and home equity will play

an important role in the old age plans of future generations Additionally, the studyshowed that cohort effects have to be considered when analyzing the trade-off effectquantitatively

The following investigation aligns with the quantitative approach of Castles andFerrera (1996) and Doling and Horsewood (2011) Instead of analyzing ownershiprates and welfare spending on a country level, we use survey data on a household level

to generate wealth measures for pension and housing equity However, we pursue thesame goal, namely, to verify the trade-off theory of Kemeny (1980) Furthermore,

as we calculate with monetary wealth measures, we are able to determine thedisplacement effect of housing wealth by pension wealth

2.3 Analytical Framework

Welfare researchers have discussed the extent to which households respond to creases in pension wealth with reductions in other wealth, although housing wealthhas not been specifically addressed Life-cycle models suggest that increasing pen-sion wealth will have a negative effect on the accumulation of other wealth (Gale,1998) Furthermore, such models predict that increases in pensions are completelyoffset by a decrease in other wealth categories The underlying assumption is thathouseholds save only for retirement However, institutional settings and individualfactors influence household behavior The real extent of the displacement effect isdifficult to assess

for German, Jones, Bevan and Quilgars (2010) for the British, Mandic (2010) for Slovenian, Naumanen and Ruonovaara (2010) for Finnish, Palmans and Decker (2010) for Belgian, Perista

2.3 Analytical Framework 9

2.3.1 Theory

We base our empirical analysis on the life-cycle model formulated by Gale (1998)and further applied by Engelhardt and Kumar (2011) as well as Alessie, Angeliniand van Santen (2013) So far, this framework has never been applied to test therelationship between housing wealth and pension wealth The model assumptions

earnings, and occupational as well as public pension benefits

Over the earning years, non-pension wealth (W ) of the household at age A equals the present value of all accumulated earnings (E) less consumption (C),

wealth equation, we rely on a constrained lifetime utility maximization framework

We assume for that purpose a constant relative risk aversion (CRRA) utility function,

which depends on (i) the present value of the sum of earnings and pension income,

but not on their allocation, and (ii) on consumption growth (x), which is defined as

x = r − δ ρ − r. (2.7)Substituting 2.6 into 2.5 and the resulting equation back into 2.1 leads to

by lifetime earnings and pension benefits In particular, equation 2.11 relates

of retirement to the time of death (T ).

2.3.2 Econometric Specification

The econometric model follows the specification of Alessie et al (2013), but relates

pension wealth to housing wealth (H) including other forms of wealth as control

variables This assumption reflects that households use their main residence as asource of income in old age to supplement pensions Based on our model equation2.11, we define the empirical model as

where

date, corrected by the adjustment factor 1 − Q, henceforth referred to as lifetime

income, and the present value of future household earnings until retirement, corrected

the disturbance term

by pension wealth We refer to it as the “trade-off” parameter

2.4 Data

provides survey data on multiple disciplines as health status, economic situation,and social networks of individuals aged 50+ in Europe It consists of six waves

2008/9, Wave 4 in 2010/11, Wave 5 in 2013 and Wave 6 in 2015 With advancingwaves similar interviews are conducted with previously questioned households, butwith more countries added and an extended range of subjects In its fourth wave,

measurements are of special interest Given that SHARE focuses on individuals aged50+, measurements tend to capture individual portfolios at the peak of the assetaccumulation curve, assuming households build up their asset wealth during theirworking age and spend their wealth during their retirement

histories are of interest Questions included not only the current situation but the history of situations.

Hungary, Ireland, Israel, Italy, Netherlands, Poland, Portugal, Sweden, Slovenia, Spain, and

We use survey responses of Wave 4 to compute wealth variables in present valueswith the base year 2010 As we limit the analysis to data obtained from just onewave and a particular time (2010/11), we end up with a cross section data set To

labor market status as well as labor income and wages of all respondents throughouttheir lives The overall analysis is limited to the intersection of countries from Wave

Parameter calculations were conducted as follows:

Brugiavini, Cavapozzi, Pasini and Trevisan (2013) and Antonova, Aranda, Pasini and Trevisan (2014) for methodological details The Job Episodes Panel release 6.0.0 is based on SHARE Waves 1, 2 and 3 (SHARELIFE) (DOIs: 10.6103/SHARE.w1.600, 10.6103/SHARE.w2.600, 10.6103/SHARE.w3.600).

Respondents got confused between the old and the new Zloty during the period of devaluation

public unemployment benefits with 80% of the last earnings We convert historicalvalues into local currencies of 2010 using the country-specific consumer price index

of the year the earning was received Subsequently, we convert local currenciesinto PPP-adjusted Euros We exclude individuals with only one wage point.Additionally, to end up with a homogeneous sample, we trim lifetime income andpension wealth of a household by 5% from above and below for each country

assump-on individuals aged 50+, assump-only a limited proportiassump-on of individuals will receive laborincome in their future

Explanatory variables are included in the regression to explain asset accumulationdifferences among households and countries As we do not use total wealth asdependent variable, we include other forms of wealth alongside pension and hous-ing wealth as controls These include net financial wealth, net business wealth,cars, and other real estate All wealth figures are expressed in Euro and arePPP-adjusted to address differences in the purchasing power of the analyzedcountries Furthermore, country dummies are included to account for differences

in pension systems across countries and for differences in housing policies that canprompt households to accumulate more housing wealth or other forms of wealth.Dummy variables for age groups account for possible cohort effects Cohorts maydiffer in their savings behavior and in their attitude towards saving through theaccumulation of housing wealth Additionally, we include interaction terms forcountry and age group dummies Another dummy variable accounts for the own-ership status of the household’s residence Households will only possess housingwealth if they are the owner of their main residence If a household rents themain residence, housing wealth will be zero A dummy variable for a partner

next known value of income, depending on which is nearer in terms of age When previous and

within the household accounts for a possible second earner within the household.

We also use a 0/1 dummy variable for children that are raised by the household;this is to account for the fact that households with children may have a bequestmotive to accumulate housing wealth Inheritances received are captured with aseparate variable to account for wealth that the household did not accumulatethrough its lifetime income but through family transfers Additionally, years

of education, including higher education, are added as a regression variable toaccount for differences in the income level

Figure 2.1 shows country averages of the household wealth allocation Includingpension wealth in a household’s asset allocation might be uncommon but accountsfor all wealth components that the average household can draw from in old age Weobserve that pension wealth and housing wealth are the two main components of totalwealth Pension wealth accounts for about 50% of the household asset allocation,with the highest value in Austria (60.0%) and the lowest value in Spain (39.2%).Net housing wealth amounts to about 30% of a household’s asset allocation Theproportion of housing wealth is highest in Spain (44.5%) and Italy (40.3%) andlowest in Switzerland (20.7%) and Sweden (22.9%) compared to the other countries.Financial assets play a certain role mainly in Western European countries In Easternand Southern European countries they represent only a small portion of the householdportfolio All other household assets account for less than 10% These include otherreal estate, cars, and business wealth

Table A.1 summarizes country averages of our non-wealth parameters The incomefigures are divided into lifetime and future income depending on the age of thehousehold When we add these two income sources, we see that households inSwitzerland and Belgium earn the most during their working years, Spanish, Czechand Italian households the least Around two thirds of the observed households arehomeowners, with the highest ownership rates in Spain and Italy and the lowest inSweden and Switzerland In the majority of cases households are occupying a house

as a couple and raised children We see low rates of major inheritances received inAustria, Czech Republic, Italy, and Spain and high rates in the Netherlands Finally,

we observe an average education period of about 11 years with lower than averageperiods in Spain, Austria, Italy, and Denmark

2.5 Regression Results

When analyzing SHARE survey data, we have to consider measurement errors forwealth, income, and pension variables First, respondents may value their real estateincorrectly When asked how much they would receive if they sold their propertytoday, we have to assume that non-experts will not report the value correctly Second,the measurement of the respondents wage path may be biased since we interpolatelifetime income and extrapolate future income Furthermore, since pension wealthand lifetime or future income are all functions of the respondents’ past income,

2.5 Regression Results 15

52.3 33.9 49.1 29.4 48.8 20.7 48.1 25.3 48.4 27.3 44.9 40.3 39.2 44.5 53.9 23.3 48.5 22.9 55.2 27.2 60.0 25.9

Business Wealth

Figure 2.1: Asset allocation of European households

Source: Own calculations based on SHARE wave 4

Notes: Figures present PPP-adjusted net figures.

measurement problems we follow the model adaptation of Alessie et al (2013) We

Since we restrict our analysis on housing wealth, this assumption is even stronger

If there is full displacement between housing and pension wealth, the parameterrestriction will induce an upward bias for the ordinary least squares (OLS) estimate of

uncorrelated

For the non-retired, pension wealth is calculated using the expected replacement rateand the current income In this case, income and pension wealth are likely correlated.Whereas for pensioners, pension wealth is calculated as the present value of future

this subgroup based on the age of the household head rather than by its retirementstatus This avoids a potential endogeneity problem associated with the retirementdecision We form the subsample with all household heads at the age of 63 or older

We pick 63 years as the cut-off point because this is the average retirement age forthe sample within all countries

We estimate the trade-off parameter on the whole dataset (Full Sample) using tionally robust and median regression techniques as in Gale (1998) and Alessie et al.(2013) to limit the impact of outliers Furthermore, we apply the same regression onthe retiree subsample (Retiree Sample) To deal with missing values, the SHAREdataset provides an imputed dataset as described by Christelis (2011) In particular,the dataset was imputed by country and by household characteristics Hence, we usemultiple imputation regression techniques according to the data structure provided.Our results are measured at household level and not at individual level becausehousing wealth is mostly accumulated by the household and not the individual Wecontrol for other forms of wealth, homeownership status, partners in the household,children, inheritances, and education as described in 2.3.2 Furthermore, we use

Table 2.1: Regression coefficients: European sample

statistically different from zero in all subsamples Hence, the trade-off effect could

be verified in our European sample of households within SHARE As the theorysuggests, our results show a negative relationship between housing and pensionwealth in the household portfolio of the elderly, i.e., households with a higher nominalamount of pension wealth possess less housing wealth than households with a smaller

homeownership (Owner) and partners in the household (Couple) significantly affecthousing wealth This is consistent across all samples and estimation methods.Since we calculate with PPP-adjusted wealth measures in Euro present values, thedisplacement effect can be quantified For the overall sample (Full Sample) thedisplacement of housing wealth by pension wealth amounts to around 6% to 10%,depending on the regression technique used Robust and median regression tech-niques produce displacement estimates that are larger in absolute terms; this isconsistent with the results of Alessie et al (2013) and Gale (1998) However, we

country-specific characteristics Age group dummies account for differences in savings behavior of different age cohorts Interaction terms of country dummies and age group dummies account for country-specific differences in cohorts.

9

2.5 Regression Results 17

may be correlated for non-retirees The true displacement effect may therefore beunderestimated

For the retiree subsample (Retiree Sample) the calculated displacement is 16% to22% Again, robust and median regressions show higher parameter values We expectthe displacement calculation of the retiree sample to be more accurate since we limitestimation biases

Delfani et al (2014) argued that a successful trade-off might only occur within acertain institutional setting In countries with less state interventions where marketallocation determines a household’s wealth allocation, a trade-off might be morelikely Therefore, we extent our analysis to investigate the trade-off on a countrylevel We implement interaction terms of pension wealth and the respective countrydummy to obtain country specific trade-off parameters The results of the respective

We can verify a country level trade-off in five of eleven observed countries regardingthe retiree sample: Germany, Sweden, France, Switzerland, and Czech Republic.All other countries show no trade-off from a within perspective The size of thedisplacement effect is larger than in the European sample The highest displacementeffect can be observed for Czech Republic (74% to 81%), the lowest for Germany(31% to 37%) Full sample estimates show lower values but the trade-off persists.According to Delfani et al (2014), Germany and France are countries with highlyregulated pension and housing markets They suggested that in these countries atrade-off is less likely Sweden and Switzerland already encourage private asset-basedpensions Here, households are likely to save into homeownership despite adequatepensions which are not exposed to market forces

Hence, our country level estimations are not entirely consistent with the hypothesissuggested by Delfani et al (2014) We could verify a trade-off between housing andpension wealth even in highly regulated markets However, the parameter estimationshave to be regarded with caution since our sample size is limited on a country level.Since all households within a country are exposed to the same institutional setting

of housing and pension markets, they are expected to pursue the same allocation,disregarding personal attitudes With a high portion of mandatory pension, theallocation of pension wealth is not an outright decision of the household Therefore,

we suggest to enhance the investigation on a more comprehensive dataset includingcountries with more liberal markets

10

Table 2.2: Regression coefficients: single country

2.6 Conclusion 19

2.6 Conclusion

Housing and pension wealth are the two main components of a household’s assetallocation Both forms of wealth can be used as a source of income in old age There-fore, both can be viewed as a way to redistribute income from working age to old age.Homeowners are able to use housing wealth as cash income by releasing housing equitythrough a sale or as income in kind by enjoying low housing costs in their old age.Hence, housing wealth serves as an income source similar to a public pension House-holds that consider the future of the public pension system to be uncertain or fear areduction of their living standards in old age despite any public pension system mayconsider housing wealth as an alternative way to provide for their old age This ideaforms the basis of the theory of a trade-off between pension claims and housing wealth.The theory assumes that households in countries with a generous welfare systemenjoy the protection of the welfare state and do not typically see any need for building

a safety net by accumulating housing wealth They rely on their public pensionprovisions in old age In contrast, households in countries with negligible publicpension systems accumulate housing wealth over their working years and use that as

a source of income in old age, either by saving rental payments or by trading downhouses and cashing in on housing equity Hence, a negative relationship betweenhousing and pension wealth is expected by the theory

From a policy viewpoint – with pressurized pension systems in most of the Europeancountries due to an older population – a change in a household’s wealth allocation can

be predicted when answering the question: How do households react to decreasingpension wealth? The trade-off theory gives one possible explanation The householdwill save into housing wealth to build a safety net for the old age They could usehousing wealth as a source of income after retirement

This chapter examined empirically the possible trade-off between housing and pensionwealth implied by the above theory The analysis made use of cross-sectional data

of households from eleven European countries as provided by SHARE We wereable to calculate present values of pension and housing wealth measures as well aslifetime and future income In our dataset, pension wealth accounts for around 50%

of the household asset allocation, while net housing wealth amounts for around 30%.Hence, both forms of wealth represent the two main assets of an household in allobserved countries We applied a regression framework based on a life-cycle modelwhich is used by welfare researchers to test the displacement effect of pension wealth

on total wealth We found evidence for a negative relationship between pensionwealth and housing wealth Furthermore, we estimated the displacement effect ofhousing wealth by pension wealth between 16% and 22% Additionally, we testedthis relationship on a country level We could verify a trade-off for Germany, Sweden,France, Switzerland, and the Czech Republic Displacement effects are at a higherlevel than in a European context But the estimated parameters have to be regarded

with caution Although our approach limits the effect of correlated measurementerrors, our dataset is limited on a country level Further research is needed on amore extensive dataset which allows to test for wider subsamples.

3 Do Europeans Release Their

Housing Wealth in Old Age?

3.1 Introduction

Growing pressure on the public pension systems of European countries induced

by an increasing life expectancy, changes in the age distribution of the population,the indebtedness of many European public households, as well as the current lowinterest environment suggests a need for a change in the public pension policies Theencouragement of an asset-based private pension provision is already a policy goal ofmost European countries A high level of private household wealth enables elderly tofinance a significant portion of their old age expenses by liquidating accumulatedassets

When approaching retirement, the portfolio of most European households consists oftwo main assets: Pension wealth and housing wealth for those who are homeowners.The majority of pension wealth is public pension wealth Public pensions are paid

by the state according to its pension system Normally, a household is uncertainabout the amount it can expect to receive, but this uncertainty declines as retirementapproaches Accumulating housing wealth follows the same life-cycle redistributionprocess of income from working age to retirement as pension wealth does But ahousehold can be more certain about the income provided by housing in the future.Income derived from housing wealth can serve as a form of pension But even infinancial distress households seem reluctant to release housing wealth (Angelini,Brugiavini & Weber, 2009) This dichotomy forms an excellent field of research whichhas to be evaluated further

For households, housing can be both a consumption good and an investment good

It provides two forms of income: Income in kind and income in cash (Doling &Horsewood, 2011) Income in kind is provided by consuming housing services whenoccupying the house It is measured as the rental value minus housing costs Income

in cash is released when selling the house, representing the value of the asset at

a certain time Assuming increasing house prices rather than falling prices over alifetime period, the net value of the housing asset increases with repayment of debtand value appreciation Therefore, housing acts as a store of wealth which can bereleased in the form of income in cash when selling the house Furthermore, financialproducts such as reverse mortgages release housing equity without selling the home,but generate additional income in old age

© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2019

T Müller, Essays on Housing and Pensions, Essays in Real Estate Research 15,

https://doi.org/10.1007/978-3-658-24955-7_3

In this chapter, we examine if homeowners use their house as a source of income

in cash by trading down or selling the house in old age Our analysis is based onSHARE With currently six waves, the survey makes it possible to study wealthchanges as the population ages Because of repeated measures through several waves,

we are able to construct a panel dataset of individual characteristics Previous studiesare mostly based on cross section data With a panel dataset we are able to controlfor unobserved heterogeneity

Furthermore, in contrast to previous work, we refrain from constructing an age-wealthpattern for old age, but instead consider the retirement age as a starting point This

is reasonable as individuals retire at different ages We identify a negative relationshipbetween the time that passed after retirement and the status of being a homeownerfor all countries analyzed in this study Moreover, by constructing country patternswith respect to the probability of being a homeowner after retirement, we see differingpatterns across countries, with some showing a stronger decrease than others Weargue that country and individual differences exist because of three main reasons:(i) socio-political reasons, which depend on the generosity of the welfare state, (ii)economic reasons, as tied to the development of real estate, mortgage and rentalmarkets, and (iii) reasons related to individual attitudes, as they pertain to howhousing is perceived, whether as part of the family’s safety net or in terms of bequestmotives

The chapter is structured as follows First, we outline the theoretical background andthe evidence on housing wealth decumulation that has been collected to date Second,

we present the empirical framework which covers the econometric specification,including the data used Next, the estimation results are presented and discussed

We conclude this chapter by summarizing our results and suggesting further research

3.2 Life-Cycle Theory and Housing Wealth

The first group of studies in the 1970s and 1980s made use of cross-section data fromhousehold sample surveys in different countries to determine the age-wealth relation-ship according to the life-cycle hypothesis Shorrocks (1975) examined British estatetax statistics to derive an age-wealth profile following a hump-shaped curve correctingfor compositional changes, for example due to mortality Masson (1986) studiedthe wealth-holdings of French households during the post-war period Deriving thepattern for different occupational groups, stable age effects according to the life-cycle

3.2 Life-Cycle Theory and Housing Wealth Decumulation 23

theory could only be obtained for some groups Burbidge and Robb (1985) foundsimilar evidence for different household types in Canadian microdata Jappelli (1999)estimated the average annual rate of wealth decumulation after retirement to bebetween 3 and 6 percent using the Italian survey of “Household Income and Wealth”

A key problem with all cross-section studies, however, is the need to control forcohort differences Cohorts differ in their lifetime income level, mortality, preferences,etc

Studies which addressed specifically the decumulation of housing wealth after ment are mostly based on US data Venti and Wise (1989) found no evidence for theuse of housing equity to finance consumption for the elderly The study examined

retire-US data from the Retirement History Survey (RHS) Their results suggested thathousing equity increases with age over time A liquidation is only likely after a shocklike the death of a spouse or after moving to a nursing home In later publications

on more extensive data from the Health and Retirement Study (HRS), the authorsconfirmed their results that elderly households who sell and buy a new home increasehome equity on average (Venti and Wise, 2004; Poterba, Venti and Wise, 2011).They conclude that “there is relatively little withdrawal of housing equity to purchaseother assets, to buy annuities, or to support consumption in old age”

Empirical evidence on European data is limited Chiuri and Jappelli (2008) structed repeated cross-section data out of 60 microeconomic surveys in 15 OECDmember states to inquire the pattern of homeownership rates Considering cohorteffects, they found declining homeownership rates after the age of 70 Differences

con-in ownership patterns are correlated with market regulations Angelcon-ini et al (2009)argued that European households fail to release home equity through the use ofinnovative financial products and that this failure causes financial hardship among theelderly In a later study, they showed that equity release by trading down depends onfinancial and mortgage market access and regulations Furthermore, the availability

of public housing and long-term care accommodations represent influencing meters (Angelini & Laferrère, 2011) But the authors noted that in most Europeancountries the necessary financial products are unavailable or uncommon With lesswell-developed mortgage markets, fewer homeowners trade down by selling even whenconfronted with financial distress Romiti and Rossi (2012) asked a slightly differentquestion: How is financial literacy influencing housing wealth decumulation? Theirfindings suggest that elderly households do not decumulate housing wealth afterretirement Furthermore, financial literacy decreases the allocation of housing wealth

para-in the household portfolio para-in old age Generally, they state that more financiallysophisticated households tend to have a more equally weighted portfolio allocationwith a lower share of illiquid assets

Our analysis uses SHARE data as previous studies (Angelini et al., 2009; Angeliniand Laferrère, 2011) but not as a cross-section Since SHARE has several connected

our analysis is now based on repeated measures of individual characteristics, whichallows us to control for unobserved heterogeneity, which previous studies were notable to do

3.3 Empirical Model

The intention of the empirical model is to answer the question: Do European elderlyrelease housing equity after retirement? A natural way to proceed would be to usehousing equity as our dependent variable In SHARE, respondents are asked aboutthe value of their main residence Moreover, the level of debt is known Hence,net figures of housing equity can be calculated But various problems arise whenanalyzing the change in housing equity over time Changes in housing values canhave numerous causes, such as general house price inflation or deflation, neighbor-hood changes, changes in regulations, etc Furthermore, the individual responding

to SHARE may not be able to correctly assess the value of her main residence.Therefore, a decrease in the stated value of the main residence from one wave toanother may not indicate a release of housing equity through a sale or a downgrade

of housing, but could instead be triggered by some other factors To correct for allvalue-influencing factors of real estate is difficult within a survey structure with muchheterogeneity between properties and measurement bias Therefore, we restrict ourexamination of the release of housing equity to the question whether the respondentchanged their ownership status from being an owner of a main residence to notowning a main residence anymore Due to the panel structure of the data, thischange can occur at any of the examined four points in time If the respondent’sownership status changes from owner to non-owner, it is expected that the property

is sold Hence, our dependent variable is ownership status, which is a binary variable,and our modeling framework is the panel logit model

In selecting our independent variables we are guided by the following key question:What is the probability of a release of housing equity in the years after retirement?The answer to this question is likely tied to the years the respondent has been inretirement We control for country and cohort effects as we expect differences amongcountries observed and in investment behavior across cohorts Differences amongcountries result mainly from differences in housing and pension policies, the structure

of financial and mortgage markets, and attitudes of people toward the allocation ofhousing equity based on historical experience Cohort differences can be explained

by differences in earnings and savings behavior resulting from changes in economicand social conditions after World War II

survey was conducted Wave 3 (SHARELIFE) asks questions on the lifetime histories This

3.4 Data 25

Additional control variables include measurements of the financial situation of therespondent, including in particular the respondent’s assessment of being in financialdistress, as this may have a dominant impact on the release of housing equity Wealso include a dummy variable for children to account for potential bequest motives

If the respondent is willing to bequeath property to her children, she might not bewilling to sell Additionally, we control for education using the ISCED code for theclassification of the level of education A detailed description of variables used, theirdimension and definition is given in Table A.4

we estimate a panel logit model with the following functional form:

their housing equity after retirement Therefore, we expect a negative relationship

variable on the probability of being a homeowner

3.4 Data

The empirical analysis is based on SHARE The SHARE dataset provides surveydata on multiple disciplines as health status, economic situation, and social networks

of individuals aged 50+ in Europe It consists of six waves with Wave 1 conducted

in 2004/05, Wave 2 in 2006/07, Wave 3/SHARELIFE in 2008/9, Wave in 4 2010/11,Wave 5 in 2013, and Wave 6 in 2015 In each wave similar interviews are given wherethe same respondents are interviewed, but with more countries added over time andthe range of subjects extended

We construct a panel dataset from the data of individuals that were surveyed in

2013) Countries that joined in later waves could not be included as we need repeatedmeasures of previously interviewed respondents Most, but not all, variables arecomparable between waves Some questions were added or rephrased We use onlyvariables that are comparable between waves To deal with missing values, theSHARE dataset provides an imputed dataset as described by Christelis (2011) In

histories are of interest Questions were asked not only about the respondents’ current situation but also about their past Therefore, it had to be excluded from our analysis Furthermore, Wave 6 was not available at the time of this analysis.

particular, the dataset was imputed in five different runs, by country and by hold characteristics Hence, we use the multiple imputation regression techniquesaccording to the data structure provided.

house-Housing wealth and pension related variables are of particular interest for our analysis.Table 3.1 summarizes the characteristics of the dataset for the relevant variables ofthe first and last wave (in parentheses) by country

Table 3.1: Summary statistics

Country

Owner-ship

Housing Wealth

ment Age

Retire-Pension Period

Financial Distress

Tertiary Educa- tion

dren

Chil-in % in %

of total wealth

in years in years in % in % in %

Austria 64 (55) 49 (45) 57.83 10.02 27 (16) 19 89 Germany 67 (68) 46 (49) 60.81 7.39 20 (18) 35 91 Sweden 79 (76) 41 (42) 64.15 8.12 16 (12) 28 93 Netherlands 68 (67) 50 (44) 64.63 7.69 19 (15) 25 90 Spain 90 (92) 74 (75) 61.33 9.55 62 (47) 6 92 Italy 82 (84) 68 (71) 58.54 9.49 66 (59) 5 91 France 78 (78) 59 (58) 59.53 10.17 33 (26) 23 90 Denmark 80 (76) 49 (43) 65.90 8.40 17 (11) 38 90 Switzerland 56 (58) 36 (37) 62.97 8.41 18 (10) 10 87 Belgium 86 (81) 61 (58) 60.38 10.24 27 (19) 27 90 Total 78 (76) 56 (54) 61.38 9.15 32 (25) 22 90

Source: Own Calculations based on SHARE Wave 1 and Wave 5 (values in parentheses)

Homeownership rates differ between the European countries examined in this study.This is the result of a number of factors, such as housing policies, welfare spend-ing, and general perceptions of housing as a safety net in old age In our survey,homeownership rates in Spain, Belgium, and Italy are the highest, with rates greater

3.4 Data 27

than 82% The lowest rates can be found in Switzerland, Austria, and Germany,with values less than 67% The average homeownership rate is 78% in the first wave

Housing wealth is a central element of the household portfolio As ownership ratesindicate, Italian and Spanish individuals are particularly “housing rich”, with housing

of Swiss and Swedish individuals is relatively small at about 40% Overall, housingwealth makes up around 56% of individual net worth in the first wave and 54% in thelast wave The share of housing wealth is crucial in the decision for homeowners tosell or not to sell housing assets If more liquid assets can be sold first, the individualmight be more reluctant to sell her home first

In terms of pension characteristics, we look at the retirement age and the years therespondent is already on a pension The retirement age in our sample varies amongcountries Respondents in Austria, Italy, and France retire on average before the age

of 60 Respondents in Denmark, the Netherlands, and Sweden work on average totheir mid-60s before they retire In our sample, people are on average around 9 years

on a pension when interviewed in Wave 1

Individuals can have various reasons to sell their main residence in old age Onereason to sell the home and release housing equity can be financial distress Around32% of all respondents stated that they have problems to make ends meet In Spainand Italy more than 60% of all respondents answered the question if they sufferfinancial distress with ’yes’ The lowest share of financially distressed respondentscan be found in Sweden, Denmark, Switzerland, and the Netherlands, with shares ofless than 20% One reason to hang on to one’s own home could be a bequest motivewhen the respondent has children 90% stated that they have one or more children.This figure is almost the same in all countries

The education level might also influence the decision to sell the home in retirement

They possessed generally a certificate, diploma, or some other academic degree Thelowest share of respondents with a tertiary education level can be found in Italy andSpain, with shares of 5% and 6%, respectively Denmark and Germany have thehighest shares, with 38% and 35%

of these European countries But since the SHARE dataset focuses on people aged 50+, the calculated homeownership rates are not comparable to those of EUROSTAT.

other real estate We did not include pension wealth in the assessment.

6